The kinetics of the mercury(II) catalyzed isomerization and aquation of the thiocyanatopentaamminecobalt(III) ion

Znorganica Chimica Acta, 15 (1975) 185-190 0 Elsevier Sequoia S.A., Lausanne - Printed in Switzerland

185

The Kinetics of the Mercury(I1) Catalyzed Isomerization of the Thiocyanatopentaamminecohalt(II1) A. ADEGITE,lb

M. ORHANOVI~,lC

Chemistry Department, Brookhaven

and Aquation

Ionla

and N. SUTIN

National Laboratory,

Upton, New York II 973, U.S.A.

Received March 27, 1975

Mercury(H) reacts with the Co(NHJ,SCNZf ion to form Co(NHJ,(SCN)Hg4+ which undergoes aquation and isomerization to yield Co(NHJfi203’ and Co(NH3),NCSHg4’. The stability constant of the binuclear complex is 1.2 M-’ and the corresponding AH, and AS, values are -5.2 kcal mol-’ and -17 cal mol-’ K-‘, respectively, at p = I.OOM and 25” C. Rate parameters k = 0.23 s-’ at 25” C, AH+ = 21.4 kcal mot’, and AS’ = 10 cal mol-’ K-’ were obtained for the aquation and isomerization of the binuclear complex. The two cobalt(III) products are formed at approximately equal rates. The mechanism of the reaction is discussed in terms of the formation of a common intermediate in the isomerization/aquation processes.

Introduction In acidic aqueous solution the Co(NH,),SCN2+ ion undergoes spontaneous isomerization to the more stable nitrogen-bonded isomer, Co(NH3),NCS2+, with a yield of >98%.’ We find that this process is catalyzed by mercury(II), but the catalyzed isomerization is accompanied by a significant aquation path to form the 3+ ion. The kinetics of the analogous Co(NH,),H,O process for the Cr(H20),SCN2+ ion have been studied previously.3 A two-stage mechanism for the mercury(I1) catalyzed reaction was proposed involving a fast equilibrium between the reactants to form the Co(H20),(SCN)Hg4+ complex* (K), followed by the rate determining aquation and isomerization of this binuclear complex (k). There was not sufficient buildup of the complex under the conditions employed in the earlier study to allow resolution of the composite second-order rate constant kK into the stability constant K and the rate constant k. Here we report a kinetic study of the mercury(I1) catalyzed aquation and isomerization of the Co(NH,), SCNZf ion: * Note that the formulations Cr(H20)5(SCN)Hg4f and Co(NH,),(SCN)Hg“+ have only stoichiometric and not structural significance.

Co(NH3),SCN2+

K + Hg2+ = Co(NH3),(SCN)Hg4+

Co(NH3),NCSHg4+ k i:

CO(NH~)~H,O~+

+ Hg(I1)

The build-up of the binuclear complex in this reaction is high enough to permit the determination of both K and k.

Experimental Materials Crude [Co(NH,),NCS](SCN), was prepared by the method of Carlin and Edwards.4 The material obtained contained a significant amount of +1 and 0 charged species (probably the bis- and tris-isothiocyanato complexes) which were not readily removed by repeated recrystallizations from dilute HC104. The crude material was therefore purified on a column of BioRad AGSOW-X2 cation exchange resin. The poorly soluble complex readily precipitated on the column when elution with NaCIO,, HC104, NaCl, or HCl L 0.6M was attempted. The complex was therefore eluted with 0.2M HCI, the eluate concentrated on a rotary evaporator at 253O”C, and the [Co(NH3),NCS]C12 precipitated by cooling the solution to 0” C. The cobalt(II1) content was determined by flame photometry or as cobalt(I1) by the Kitson method’ after reduction of the cobalt(II1) by europium(I1). Thiocyanate in the reduced solution was determined spectrophotometritally as FeNCS’+ after the addition of excess iron(III).6 The [Co(NH,),NCS]C12 prepared in the above manner had molar absorptivities of 178.1 and 1541M-’ cm-’ at 498 and 306 nm, the absorption band maxima, in very good agreement with absorptivities of 177.4 and 15.5OM-’ cm-’ at 497 and 305 nm, respectively, reported by Schug et al.’ for the complex prepared in a different way. The literature controversy concerning the molar absorptivities of the 2+ ion cited by Schug’ thus appears to Co(NH,),NCS be resolved.

A. Adegite, M. OrhanoviC and N. Sutin

186

The sulfur-bonded isomer [Co(NH,),SCN]Cl,. 1.5 H,O was prepared by the literature method.’ The purity of the complex was ascertained from absorption measurements in aqueous solution. The amount of the Co(NH,),NCS*+ component (O-3%) present as a preparative impurity and formed by the isomerization process was calculated from absorption measurements at 512 and 288 nm using molar absorptivity values of 74 and 15,6OOM-’ cm-’ for CO(NH,),SCN~+,~ and 161.4 and 1232M-’ cm-’ for Co(NH,),NCS2+, respectively. [CO(NH,),H,O](CIO,), was prepared using a previously described procedure’ and ,its purity checked spectrophotometrically.8 The other chemicals were all reagent grade and were used without further purification. Triply distilled water was used throughout. Kinetic

M-l

cm-‘, respectively. If free thiocyanate is present in solution as the counter ion of the complex and/or as preparative impurity, the maximum at 336 nm is obscured by the absorption of mercury(II)-thiocyanate species. The relative amounts of Co(NH,),NCS’+ and Co (NH3)5H203+ produced in the mercury(I1) induced reaction were calculated from the absorption measurements at 476 nm using F = 45.6 and 85.2M-1 cm-’ for the Co(NH3),H203+ and Co(NH,),NCS’+ ions. respectively, in the media employed.

Results Kinetics

In the presence of excess Hg(CIO,), the rate disappearance of Co(NH3),SCN2+ is given by

Measurements

The kinetics were followed at 288 nm, an absorbance maximum for Co(NH,),SCN’+. A Cary 14 instrument which was equipped with a thermostated cell compartment fitted with a magnetic stirrer was used for the measurements. All of the reaction components except the cobalt(II1) were placed in a modified 10 cm cell containing a stirring bar, the cell was thermostated to the reaction temperature, and the reaction was initiated by injecting 1 ml of a [Co(NH,),SCN]C12 solution into the stirred reaction mixture. The cobalt(II1) concentration in the runs was (S-7) x l(T’M, and the concentration of HgjCIO,), was high enough so as to remain constant during a run. The pseudo-first-order rate constants, kobsd, for the runs were calculated from the straight lines obtained by plotting log(A,A,) vs. t, where A, and A, are the absorbances at time t and after 10 half-lives of the reaction, respectively. The ionic strength was maintained at 1 .OOM by the addition of Mg(C104)Z or NaCIO,.

[Co(NH3)sSCN2+1 = kobsd[Co(NH

-d

dt

Analysis

) SCN2+]

35

Figure 1 is a plot of kobsd (25” C, ionic strength 1 .OOM with sodium perchlorate) as a function of mercury(I1) concentration. The broken line shown there (slope 0.30M-’ set-I) was drawn on the basis of the first five entries in Table I (which are not shown in Figure 1). These lie in the region of low mercury(I1) concentration (2.26 X IO4 to 6.67 x 1F3M) in which k obsd is linearly dependent on [Hg(II)]. Significant departure from linearity is observed in Figure 1 at the higher mercury(I1) concentrations. It is not likely that the high mercury(I1) behavior is due to a medium effect, as the same kind of behavior is found with magnesium perchlorate as supporting electrolyte (in Table I, note the diminishing value of k,,,,/[Hg(II)] as [Hg(II)] increases).

I

Product

of

I

I

I

ITI

I

I

6-

The Co(NH,),NCS2+ and Co(NH,),H,03+ produced in the reaction were determined spectrophotometrically using a 10 cm cell. For this purpose runs with (S-7) x lO4M [Co(NH,),SCN]CI, and 0.1 or O.OlM were performed. The spectrum of the Hg(ClO& Co(NH3),NCS2+ product depends on the mercury(I1) concentration.’ For runs with 0.01 M Hg(I1) the concentration of mercury(II) in the “infinite time” solution was increased to 0.1 M prior to the spectrophotometric measurements to ensure complete formation of the Co(NH3),NCSHg4+ adduct.” This binuclear complex is very inert, the half-life for its aquation being >lO days. Its visible-near UV spectrum has not been well-defined previously.9,‘0 In a solution containing 0.1 M HClO,, 0.2 M Mg(C104)2, and 0.1 M Hg(ClO,), we find absorbance maxima at 476 and 336 nm with molar absorptivities of 85.2 and 79.0

IIIIIIIIIIIILll 0.1

02

0.3

[HgW],M

Figure 1. Pseudo-first-order rate constants for the mercury(I1) catalyzed aquation and isomerization of Co(NH,),SCN*+ as a function of Hg(CIO,), concentration in O.lM HCIO, andp = l.OOM (NaClO,) at 25°C.

Hg(ll)

Catalyzed Reactions of Co(NHJ,

SCN2+

187

TABLE I. Rate Constants for the Mercury(H) Aquation and Isomerization of Co(NH,),SCN*+ O.lM HCIO, andp = l.OMa,b.

Catalyzed Ion at

and

T “C

(Hg’+) mM

25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 25.0 38.6 38.6 38.6 38.6 38.6 38.6 38.6 38.6 38.6 10.85 10.85 10.85 10.85 10.85 10.85 10.85 10.85 10.85 10.85 10.85 10.85 10.85

kobsd x 103 set-’

0.226 0.667 2.26 2.26 6.67 0.361 0.45 1 0.677 1.58 4.51 14.9 35.0 79.6 125 157 187 215 260 300 0.150 0.350 0.702 1.60 5.00 15.0 35.1 79.8 100 1.49 1.99 3.02 4.43 9.91 19.9 49.5 80.2 120 171 220 260 299

10 x k,,sa/(Hg*+) M-’ see-’

0.0671’ 0.217’ 0.704c 0.72c,d 2.10’ 0.105 0.130 0.188 0.458 1.26 4.20 9.06 20.7 31.4 37.2 43.3 49.8 55.5 62.1 0.142 0.324 0.653 1.51 4.62 13.9 30.4 65 93 0.099 0.131 0.197 0.295 0.653 1.31 3.01 4.62 6.61 8.4 10.5 11.9 13.5

These

observations

scheme Co(NH,),SCN’+

step

are

consistent

with

K + Hg’+ + Co(NH,),(SCN)Hg4+--%products

the first step is a fast equilibrium is rate

is given by

determining.

In terms

the

its inverse

’

by

+ l/k

On plotting the data in Figure 1 and Table I as l/kobsd vs. l/[Hg*+] straight lines are obtained. Figure 2 shows the plot for the Mg(ClO,), medium at 25” C. Values of K and k calculated from the slopes and intercepts of the plots are given in Table II. The lower Hg’+ concentrations used (and the lower K) at 38.6” C precluded resolution of the kK product. From the plot of log(kK/T) vs. l/T for the three temperatures in Table II the following values for the composite reaction parameters were calculated: (AHi + AH,) = 16.2 kcal mol-’ and (AS* + AS,) = -6.7 cal mol-’ K-‘. Using the analogous plot, log(k/T) vs. l/T for the two k entries, activation parameters AH1 = 21.4 kcal mol-’ and AS+ = 10 cal mol-’ K-’ were obtained. The standard enthalpy and entropy changes for the formation of the binuclear complex are then AH, = -5.2 kcal mol-’ and AS, = -17 cal mol-’ K-‘.

2.97’ 3.21’ 3.12’ 3.19c,d 3.10’ 2.91 2.87 2.78 2.90 2.80 2.82 2.59 2.60 2.51 2.37 2.31 2.32 2.14 2.07 9.45 9.25 9.3 9.4 9.25 9.3 8.65 8.2 8.25 0.665 0.66 0.655 0.665 0.66 0.66 0.61 0.575 0.55 0.49 0.475 0.46 0.45

-

kK[Hg’+] l+K[HgZ+]

l/k obsd = l/kK[Hg’+]

Figure 2. Plot of k,,,,l as a function of [Hg(II)]’ for the and mercury(I1) catalyzed aquation isomerization of Co(NH8)sSCNZ’ in 0.1 M HC104 andp = l.OOM (Mg(ClO,),) at 25°C.

a [Co(NH,),SCN]*+ = 5-7 x 10-6M. b Ionic strength adjusted with Mg(CIO& unless otherwise noted. ’ Ionic strength adjusted with NaCIO,. d HClO* = 1.24 x 10m3M.

where

k obsd =

reaction

(1)

and the second scheme kobsd

TABLE II. Rate and Equilibrium Constants for the Reaction between Co(NH3),SCN*+ and Hg*+ at HC104 = O.lM and I( = l.OM. T “C

kK M-’ set-’

k set-’

K M-1

10.85’ 25.0” 38.6” 25.0b

0.066 0.285 0.93 0.30

0.036 0.23 -

1.8 1.2 -

0.23

1.3

of this

a Mg(CIO,),

medium.

’ NaCIO,

medium.

A. Adegite, M. OrhanoviC and N. &tin

188 Cobalt(IlI)

Reaction

Products

In order to perform the spectrophotometric analysis of the cobalt(II1) reaction products it was necessary to resolve the literature controversy’,” regarding the visible-near UV absorption spectrum of Co(NH,), NCSHg4+. In the presence of 0.1 M Hg(CIO,)Z carefully purified [Co(NH,),NCS]CI, shows absorption maxima at 476 and 336 nm, the two maxima originally reported by Larsson.’ The existence of the maximum at 336 nm is not unexpected. The maxima in the Co(NH,),NCS*+ spectrum are lowered in intensity and displaced toward higher energies in the spectrum of Co(NH,),NCSHg4+ with the shoulder of the former ion at -360 nm appearing in the spectrum of the binuclear complex as a maximum at 336 nm. The spectrum of the cobalt(II1) products of the mercury(I1) catalyzed reaction indicates the formation of both Co(NH,),NCS’+ and Co(NH,),H,O”+. At 65, 38.6, 25.0, and 10.4”C the amounts of the Co(NH,),NCS *+ isomer formed are 59, 39, 47, and SO percent of the total cobalt(II1). A relatively large experimental error introduces an uncertainty of about -+4% in these figures. This essentially precludes the drawing of conclusions about a possible trend in the product distribution with temperature.

Discussion Equilibrium constants for some mercury(I1) reactions relevant to this study are collected in Table 111. It is evident from this table that the equilibrium constants for mercury(I1) addition to the coordinated thiocyanate in the nitrogen-bonded complexes Co(NH,),NCS2+ and Cr(H,O),NCS’+ are factors of -lo4 and 7 x 104, respectively, less than the equilibrium constant for mercury(I1) addition to free thiocyanate. In these reactions the mercury(I1) presumably adds to the sulfur atom of the thiocyanate. It is also evident from Table III that the equilibrium constants for mercury(I1) addition to the sulfur-bonded complexes Co(NH,),SCN*+ and Cr(H20),SCN2+ are reduced by an additional factor of -10’ and > lo4

TABLE at 25°C

III. Stability Constants andp = IM.

for Reactions

of Mercury(B)

compared to the constants for mercury(I1) addition to the corresponding nitrogen-bonded complexes. In the sulfur-bonded complexes the mercury(I1) may add to the nitrogen atom or to the sulfur atom of the coordinated thiocyanate. Consequently the reduced stabilities of the resulting binuclear complexes may reflect either the poorer affinity of Hg*+ for the nitrogen atom or steric and electronic effects related to the binding of mercury(I1) to the coordinated sulfur atom. If the affinity of Hg*+ for free azide is taken as a measure of the stability of a nitrogen-bonded HgNCS+ complex, the possibility of mercury(lI) binding to the nitrogen atom of the sulfur-bonded complexes cannot be ruled out. In a study of the mercury(I1) catalyzed aquation of Cr(H20)sN3*+ the reaction was found to be first order in mercury(II) over the range of concentrations used indicating a value of < 7 W’ at 25 a C* for the stability constant of the mercury(I1) adduct.” These comparisons show that the magnitudes of the stability constants for mercury(I1) addition to the sulfur-bonded complexes do not indicate the mode of mercury(I1) binding. Additional information can be obtained from relative rate considerations. Rate constants for the aquation and/or isomerization of cobalt(III) and chromium(II1) complexes containing coordinated thiocyanate are presented in Table IV. The catalytic effect of mercury(II) bound to the remote end of the thiocyanate on the rate of aquation of Co(NH,),NCS*+ and Cr(H,O),NCS *+ is seen to be < 2.5 x 103 and -104, respectively, compared to the rates of spontaneous aquation of the nitrogen-bonded complexes. Similarly, the catalytic effect of coordinated mercury(I1) on the aquation/isomerization rate of Co(NH,),SCN*+ and Cr(H,O),SCN *+ is 2.9 x 10’ and > 9 x 1O6 compared to the rates of spontaneous aquation/isomerization of the sulfur-bonded complexes. The much larger catalytic effect of mercury(l1) on the sulfur-bonded complexes suggests that in these complexes mercury(I1) addition

takes

* Estimated independent

place

to

the

coordinated

sulfur

from the upper limit of K at 60” C for the acidreaction and assuming d H, = -S kcal moT’.

with Some Coordinated

or Free Pseudo-Halide

Ligands

Reactant

K, M-’

Reactant

K, M-’

Co(NH&SCN’+ Co(NH,),NCS’+ SCN-

1.2” 9.8 x 104b,’ 1.2 x 109d

Cr(H,O),SCN*+ Cr(H,O),NCS*+

<1e 1.7 x 104’ 1.4 x 107g.h

N;

atom

“This work. ‘Ref. 10. ‘p = 0.1 M, K = 2.6 x 1O-5 M-’ inp = 0.16M K. Schug and B. Miniatas, Abstracts, 255th National Meeting of the American Chemical Society, San Francisco, Calif., March 1968, No. M-136. “L. Ciavatta and M. Grimaldi, Inorg. Chim. Acta, 4, 312 (1970). ‘Ref. 3. ‘Ref. 12. ST. R. Musgrave and R. H. Keller, Inorg. C/rem., 4, 1793 (196.5). “At 28°C andp = 0.25M, the value decreases with increasingp.

-

Hg(ll)

Catalyzed Reactions of Co(NHJ,

SCN2+

189

TABLE IV. Rate Constants for the Isomerization SCK Ligand at 25” C and p = 1 M.

and/or

Aquation

of Some Co(III)

and Cr(II1) Complexes

with

Reactant

k, see-’

Reactant

k, set-’

Co(NH&(SCN)H$+ Co(NH&NCSHg4+ Co(NH&SCN’+ Co(NH,),NCS’+

0.23”
Cr(H,0),(SCN)Hg4+ Cr(H20),NCSHg4+ Cr(H20)sSCN2+ Cr(H,O)sNCS’+

>5.4 8.5 x 5.9 x 9.1 x

“This work. bRef. 2. ‘D. L. Gay and G. C. Lalor,J. Chem. Sot. (A), 1179 (1966). E. L. King,J. Phys. Chem., 59, 1217 (1955). 8Ionic strength OSOM.

rather than to the “free” nitrogen atom of the thiocyanate.* Another possibility that needs to be considered is the existence of a rapidly established, mercury(II)-dependent equilibrium in addition to the one observed kinetically. This adduct would not have been detected in the kinetic measurements if the stability constant for its formation were sufficiently high. However, the visible-near UV spectrum of Co(NH,),SCN*+ does not show any change in the presence of 5 x lF3M Hg’+ at 10.9” C. Since the reaction rate is slow enough under these conditions to safely allow “zero time” absorbance measurements to be made, there is no spectrophotometric evidence for the formation of a very stable complex between Hg2+ and Co(NH,), SCN*+. It is likely that the intramolecular mechanism proposed previously23336 for the spontaneous and the mercury(I1) catalyzed rearrangement of thiocyanate complexes applies also to the isomerization of Co(NH,),SCNHg4+. This mechanism involves breaking of the metal(III)-sulfur bond, rotation of the ligand within the solvent cage, and rebonding. Assuming the existence of a common intermediate I for the isomerization and aquation of Co(NH,),(SCN)Hg4+, the reactions of this binuclear complex can be represented by the following scheme ~.___ *This conclusion must be regarded as tentative at this time since the binuclear complex formed by mercury(I1) addition to the nitrogen atom of the thiocyanate might also be relatively reactive because of the more for its aquation/isomerization.

TABLE

V. Isomerization

favorable

driving

force

Yield for Some Metal Complexes

dRef. 3. ‘Ref.

12. ‘C. Postmus

Co(NH,),(SCN)Hg4+&

x lo* d 1O-5e lo-5 d 1cr9’~*

and

%o(NH,),NCSHg4+ k1 I

I

k, Hz0 Co(NH,),H,O’+ The steady-state gives

for the concentration

k,k, k, + k, + k,

Since the aquation and isomerization paths proceed to comparable extents, it follows that k, = k3. The relative values of k, and k, are determined by at least three factors: the extent of x-bonding in the intermediate, the cage effect, and the charge of the rotating (or leaving) group. We comment on these on the basis of the data presented in Table V. Comparison of the percentage isomerization of and Co(NH,),ONOZt suggests Co(NH,),SCN’+ that n-bonding of the rotating group in the intermediate may not be required for high isomerization yields. A similar conclusion concerning the role of x-bonding in the isomerization of Ru(NHJ5N,‘+ has been reached These considerations imply that the previously.‘3 solvent cage rather than n-bonding is primarily responsible for kz/k3 > 1 in these systems. The relatively low percentage isomerization of Cr(H,O),SCN” compared with the spontaneous isomerizations of the three

in Aqueous

Solution.

Complex

% Isomerization

Complex

% Isomerization

Co(NH,),SCN’+ Co(NH&ONO*+ Ru(NH&N,‘+ Cr(H20),SCN2+

>98” -98--99b -98’ 72d

Co(NH,),(SCN)Hg4+

50’

“Ref. ‘Ref.

2. bR. G. Pearson, P. M. Henry, J. G. Bergmann 13. dRef. 3. eThis work. ‘Ref. 3.

of I

k,k, k1 + k2 + k,

kisom =

k,, =

assumption

+ HgSCN+

Cr(H20),(SCN)Hg4+ and F. Basolo,

J. Am. Chem. Sot., 76, 5920 (1954);

Ref. 13.

A. Adegite, M. OrhanoviC and N. Sutin

190

other complexes shown in Table V may stem from an associative component in the activation of chromium(II1). Substitution reactions of chromium(II1) often feature an I, mechanism,14 and a bond-making role of H,O in the activation for the isomerization could lead to a higher degree of aquation. The relatively lower ratio of isomerization to aquation for Co(NH,), and Cr(H,O),(SCN)Hg”+ may reflect (SCN)Hg4+ the large size and relative ease of escape of the I+ charged HgSCN+ from the solvent cage compared with the l- charged ligand in Co(NH,),SCN’+, Cr and Co(NH,),ONO*+, or the zero(H,O),SCN*+ charged ligand in Ru(NH,),N,‘~. .On the basis of these comparisons we may conclude that in these systems the separation of the metal center and the (rotating) ligand is prevented by the surrounding solvent molecules for a time long enough for the isomerization to occur, and that, all other factors remaining constant, the isomerization/aquation ratio is higher the greater the S,l character of the initial ligand separation.

2 3 4 5 6 7 8 9 10 11 12 13

References 1 a) Research performed Energy Research and

14 under the auspices of the U.S. Development Administration. b)

Fullbright-Hayes Scholar from Chemistry Department, University of Ife, Nigeria. c) On leave of absence from Institute “Ruder Boskovic”, Zagreb, Croatia, Yugoslavia. D.A. Buckingham, 1.1. Creaser and A.M. Sargeson, Inorg. Chem., 9, 655 (1970) M. Orhanovic and N. Sutin. J. Am. Chem. Sot., 90, 4286 (1968). R.L. Carlin and J.O. Edwards, J. Inorg. Nucl. Chem., 6, 217 (19%). R.E. Kitson, Anal. Chem., 22, 664 (1950). A. Haim and N. Sutin, J. Am. Chem. Sot., 88, 434 (1966). K. Schug, M.D. Gilmore and L.A. Olson, Inorg. Chem., 6, 2180 (1967). R.C. Splinter, S. J. Harris and S. Tobias, Inorg. Chem., 7, X97 (1968). R. Larsson, Acta Chem. &and., 16, 931 (1962). L.C. Falk and R.G. Linck. Inorg. Chem., 10, 215 (1971). V. Mahalec and M. Orhanovic, Znorg. Chim. Acta, 5, 457 (1971). J.N. Armor and A. Haim, J. Am. Chem. Sot., 93, 867 (1970). J.N. Armor and H. Taube, J. Am. Chem. Sot., 92, 2560 (1970). G. Guastalla and T.W. Swaddle, Can. J. Chem., 51, 821 (1973); D. R. Stranks and T. W. Swaddle, J. Am. Chrm. Sot.. 93, 2783 (1971).